;

Initial reports indicate that Microsoft's global service outage was caused by a bug introduced several months ago. The bug was initially tested on low-spec machines, which delayed detection until the problem eventually surfaced.
Microsoft released an initial report after the incident, stating that the Microsoft 365 service outage worldwide on 12-13 Nov 2026 originated from a bug in a network gateway server update released in July. The bug remained hidden without symptoms for over three months before being triggered by a surge in usage coinciding with a drop in server capacity.
This report was published on the Microsoft 365 admin center on 15 Nov. Charles Lamanna, Executive Vice President of Copilot, later revealed on X that the hidden defect caused servers to become overworked and unresponsive under certain load conditions.
The July update was developed specifically for Microsoft 365 network gateway servers. It underwent load testing and had been used globally without issues. However, the defect caused lock contention within a Windows process, exhausting the server's CPU and memory resources.
Microsoft stated that testing did not reveal the problem because it focused on hardware with fewer cores and lower specs, while the issue worsened on newer machines with more cores. Besides the bug, two other factors occurred simultaneously: first, Teams failed over to using the same gateway as Exchange Online, increasing the load; second, an unrelated issue from the previous day caused the gateway server capacity to drop by nearly 14 percent.
When one server stopped responding, traffic was rerouted to the next functioning server, which suffered from the same bug and also failed, causing a cascading failure.
Recovery took a long time because queued traffic hit the servers immediately upon coming back online, causing repeated crashes. Microsoft admitted that the automatic recovery system was not designed to handle failures on this large a scale.
Lamanna explained that because the root cause was a hidden bug present for months, the usual rollback mechanisms, staged feature releases, and feature flags could not rapidly reverse the changes.
Microsoft engineers wrote new updates to attempt restoring normal operation and gradually deployed them to servers in groups. Services like Teams, OneDrive, SharePoint, and Office apps were able to move off the problematic gateway and resumed faster, but Exchange Online could not due to system size constraints, making Outlook the longest-affected service.
Bloomberg reported that over 1,000 engineers and staff participated in the incident response meeting, led by Lamanna and Scott Guthrie, Microsoft's cloud executive. They confirmed the incident was a technical issue unrelated to cyberattacks or AI agents.
This outage severely impacted a wide range of organizations including government agencies, courts, hospitals, and businesses across 74 countries worldwide.
. TNW